杏吧原创

WHEN Rashied Amini split up with his girlfriend of two years, she said she鈥檇 need to see a cost-benefit analysis before she would consider taking him back. At the time, the engineer laughed off the idea. Now, a few months later, he鈥檚 about to launch , an online tool that calculates your odds of being lucky in love.

鈥淚 thought, 鈥業鈥檝e made cost-benefit models in the past, let me see if I can start doing something like this鈥�,鈥� Amini says. 鈥淚 was going through a break-up and it was something to turn my attention to.鈥�

Amini鈥檚 former job involved tweaking mission parameters to find the optimal conditions for success in NASA鈥檚 now-shelved Constellation programme, which looked at building bases on the moon or Mars. With Nanaya, he wants to help people find success in their personal lives. The idea is to forecast your probability of finding a suitable romantic match if you鈥檙e single, or if not, of finding someone who is a better match.

Amini is far from the first to tot up the pros and cons in search of insight into romance. Before Charles Darwin proposed to Emma Wedgwood in 1838, on the back of an old letter. Marriage would be time-consuming and expensive, Darwin noted, but it did offer constant companionship, 鈥渂etter than a dog anyhow鈥�. In the decades since, many have attempted to distil the messy reality of life and romance into clean formulae, from dating site OkCupid, which mines the data of its millions of users for telltale patterns, to a mathematical model from the University of Washington that can predict whether a couple will get divorced.

聯Darwin noted that the constant companionship of marriage was 鈥榖etter than a dog anyhow鈥櫬�

For Nanaya to assess your likelihood of finding a good match, you fill out a detailed questionnaire. First, there鈥檚 a personality test supplied by Traitify in Baltimore, Maryland, which pinpoints different facets of your personality and the characteristics you鈥檇 like in a partner by asking you to pick from a series of images. Another set of questions then identifies how likely you are to meet new people in your daily life. The results are fed into a simulation that randomly mixes traits across a population of potential matches, so that each member of the population is like a roll of the dice. The size of that population is dependent on how many people you鈥檙e likely to run into 鈥� the more people you might meet, the more rolls of the dice you get.

Dating data

Nanaya spits out scores that measure your selectivity, how much your social network could grow with a given mate, your overall romantic opportunity, and a graph of your prospects over the next few years. Those who are already in a relationship will also be asked a set of questions that scores them on their compatibility and happiness with their partner.

Amini tested Nanaya鈥檚 algorithm by meeting up at coffee shops with people he had recruited as volunteers for the study through ads on classifieds site Craigslist. He asked them detailed questions about their love lives and ran their data. In some ways, he says, a human life turned out to be more difficult to tackle than a space project.

鈥淚 want people to be open-minded when they鈥檙e going into it, but obviously I also don鈥檛 want people following it blindly,鈥� says Amini. 鈥淣anaya helps you to reflect on what you want in life.鈥�

Some psychologists are sceptical about Nanaya鈥檚 potential. Attraction between people is difficult to predict, says , a psychologist at the University of Rochester in New York. 鈥淵ou simply can鈥檛 do it from paper and pencil characteristics.鈥�

at the University of Texas at Austin agrees that the program is probably incomplete. 鈥淢atchmaking algorithms are probably possible in principle, but they would likely need to incorporate data assessed after, not before, two people have met each other,鈥� he says.

An early version of Nanaya will be available at the end of February, and it is likely to evolve as more data comes in from users. Amini hasn鈥檛 tried the process on himself 鈥� yet. 鈥淚鈥檒l definitely do it,鈥� he says, once he finishes developing the prototype.

This article appeared in print under the headline 鈥淟ove at first site鈥�